Download Team #119 M Challenge Fourth Place: Meritorious Team Prize of

Team #119
M3 Challenge Fourth Place: Meritorious Team Prize of $7,500
Bergen County Academies
Hackensack, New Jersey
Coach: Elizabeth Casarico
Students: Joshua Eiseman, Peter Humanik, Elan Paz Kugelmass, Taesup Lee, Jordan
Moldow
Team ID #119, Page 2 of 18
SUMMARY
In response to a recession bred by a perfect storm of financial crises, housing market
implosions, deflation, and enormous increases in unemployment, President Obama and the
Democratic-controlled Congress introduced sweeping legislation aimed at strengthening the job
market. Warning that inaction would lead to an economic downturn comparable to the Great
Depression, the White House pushed the American Recovery and Reinvestment Act of 2009
through Congress. Despite promising to “create or save” 3,000,000 jobs, numerous questions
remain as to the efficacy and timeliness of this enormous expenditure. 20% of the funds allocated in
the law will be spent this year. A further 44% will be spent in 2010. Will the act reduce
unemployment? Was the money directed to the most effective programs, sectors and agencies? Will
further stimulation of the economy be required? If so, how much money should be allocated in the
future, and which programs will be priorities? This paper attempts an ex post facto dissection of the
so-called stimulus package, investigating its potential impact on today’s key economic indicator, the
unemployment rate.
The Federal Reserve recently estimated that recovery will begin in early 2010. Therefore,
this paper examines the effects of the stimulus package over the next two years. It is generally
agreed that deficit spending is most effective while the economy is declining. Even in pessimistic
scenarios, only 64% of the $787 billion will be spent during the recession. Therefore, only planned
expenditures for the 2009 and 2010 fiscal years were considered for this analysis. The four largest
areas by expenditure were analyzed to predict changes in employment over the next two years.
Federal disbursements and revenue reductions with respect to (1) energy, (2) taxes, (3) housing and
urban development, and (4) health and education were used to estimate changes in unemployment.
Empirically, historical employment values were coupled with various economic indicators to ease
the estimation of future changes.
Computer models determined that outlays in the renewable energy sector had large impacts
on employment. This analysis predicted that employment will rise 63.5%, strengthening this
tenuous and small industry. However, it is notable that the amount of money allocated is far greater
than the size of the industry, and that this prediction is an overestimation.
The reduction in taxes mandated by the bill is a promising way to increase employment
across the workforce. Optimistic predictions show that there will be an 8% increase in employment.
The model estimated 7.8% growth in the housing sector, which will be a welcome relief for
this devastated industry. However, it will not do nearly enough to bring the housing sector back to
baseline within the year. Housing requires a one-year lag to observe visible employment increases.
The industry will not recover before 2011 at the earliest.
Despite allocating over 30% of the spending to the health and education sectors, there is a
lack of evidence for significant job growth there. Employment in these industries was found to be
largely independent of the business cycle; i.e., employment is unaffected by the recession and
growth of the rest of the economy. Therefore, it seems unlikely that the stimulus package will
stimulate major growth.
Despite the huge package authorized by Congress, it is apparent that recovery will be
minimal or completely absent in 2009. Therefore, this paper proposes a second stimulus package
focusing on corporate income taxes. By their very nature, corporate taxes reduce the amount of
money available for salaries, depressing employment. They also create additional costs for the
company, which are passed along to the consumer, creating a financial burden in a difficult
economic period. Assuming the corporate income tax was completely eliminated, the economy
would see growth approaching 5% for 2009, exceeding growth over the past decade. This would
offer a speedy and strong recovery.
Team ID #119, Page 3 of 18
Table of Contents
I. Summary
II. Introduction
III. Design
IV. Unemployment and Stimulus Package
A. Energy
a. Introduction
b. Assumptions
c. The Model
d. Conclusion
B. Taxes
a. Introduction
b. Assumptions
c. The Model
d. Conclusion
C. Housing and Urban Development
a. Introduction
b. Assumptions
c. The Model
d. Conclusion
D. Labor, Health, and Education
a. Introduction
b. Assumptions
c. The Model
d. Conclusion
V. Further Stimulus
VI. Conclusion—Timing and Testing
VII. Works Cited
Team ID #119, Page 4 of 18
II. Introduction
In order to analyze the overall efficacy of the American Recovery and Reinvestment Act, various
assumptions and simplifications had to be made. Because of the sheer number of uncontrollable
variables in the economy, it is impossible to ensure that all variables would be given full
consideration as merited.
One popular solution to this problem of the difficulty of isolating indicators and variables of interest
is to introduce extensive econometric models. These are comprised of sophisticated formulas that
hold certain variables constant and through complex statistical methodology isolate and extrapolate
economic data of interest. This approach, despite its promise with the advent of the rapid increase in
the availability of computers, the growth of the Internet, and a general abundance of technology,
yielded much disappointment within academia. Nobel laureate and famed American economist Paul
Samuelson said regarding econometric models,
Let me make a confession. Back when I was 20, I could perceive the great progress that was being
made in econometric methods. Even without foreseeing the onset of the computer age, with its
cheapening of calculations, I expected that the new econometrics would enable us to narrow down
the uncertainties of our economic theories…My expectation is that this has not worked out.
The failure of econometric models simply proves that computers are incapable of predicting the
behaviors of living, flesh-and-blood participants in the economy.
Thus, despite the probable aptness of an econometric model to answer the question of how effective
the American Recovery and Reinvestment Act will be in resuscitating employment in the US
economy, we felt that using an empirical approach to data analysis would prove to be superior to a
dependence on econometric models for two reasons. One, instead of making broad and nebulous
assumptions regarding the economy, that any econometric model calls for, an empirical approach
rooted in graphical analysis allowed us to utilize actual economic conditions. Furthermore, since our
results as well as our assumptions were taken right from reputable government and private
statistical agencies such as the Labor Bureau of Statistics, Bureau of Economic Analysis, and
Energy Information Agency, we felt our results were far more reputable themselves as well than
any result of any complex econometric model.
Nonetheless, some assumptions were made, as they inevitably must be in any analysis of the
economy. In tackling the problem of evaluating the efficacy of the American Recovery and
Reinvestment Act, we made the following assumptions:
z
The impact of the stimulus package on each individual sector of the economy, such as
energy and health, was assumed to be directly reflected in a representative indicator of that
sector. For example, in analyzing the renewable energy sector, private consumption of
solar/photovoltaic energy was assumed to reflect the impact of the stimulus spending and
investment on that industry with sufficient sensitivity to discern the impact of the stimulus
on that sector.
z
The stimulus bill would be completely effective in stimulating the economy. Though in
essence, a Keynesian multiplier of 1.5 was assumed, this number was used sparingly due
to the lack of consensus on an exact figure within the economic literature as well as the
demanding current circumstances of a global economic meltdown that deviate significantly
from any historical precedents.
Team ID #119, Page 5 of 18
z
Any major economic crises that are not foreseeable in the immediate future, excluding
eventualities such as major GDP contractions or mass failure of the banks precipitated by
debt deflation and a complete absence of credit demand, were assumed not to take place.
z
Only that portion of the stimulus package to be spent by the year 2010 was considered in
our analysis due to the lagging effect of stimulus due to impeding factors such as
implementation and a defunct banking system as well as the fact that the bulk, 69%, of the
stimulus package $526 billion is to be spent by 2010.
With these assumptions, we tracked data series of a representative indicator of key industries
affected by the stimulus package, based on the portion of the $526 billion spent by 2010 committed
to spending and investment in that industry. The sectors analyzed were energy, labor and health,
transportation, housing, and urban development. The impact of tax cuts, though not an industry, was
also analyzed due to the large portion of the bailout that plan tax cuts occupied.
These indicators were tracked, over time, with employment figures in each sector retrieved from the
Bureau of Labor Statistics. The correlations between these indicators were measured using an
algorithm to optimize coefficient of determination between the two data series by varying the time
lag between the two series (i.e., shifting the data series horizontally to maximize the correlation).
Finally, the model was used to predict an increase in employment due to the impact of fiscal
spending in that industry.
III. Design
In order to fully understand how the money from the stimulus bill might affect employment
rates in different work sectors, we needed to predict the correlation between the number of people
employed in a sector and the magnitude of direct impact of the stimulus on the sector. To make this
connection, we assumed that, one, the stimulus would be completely successful in directly
impacting the sector in question as prescribed in the content of the stimulus bill, and two, the
magnitude of this impact could be directly measured through a representative indicator of that
sector. Therefore, we were able to make a model for the effectiveness of the stimulus to generate
job growth in each of six public sectors by comparing past employment data with past sector
strength indicators. For example, for renewable energies, we assumed that the consumption of
solar/photovoltaic energy was a direct indicator of money invested in the renewable energy industry
and compared these data to employment figures in the same industry over time.
Public data on employment figures and strength indicators in each of the sectors was gathered
from the Internet. The information gave annual figures, and from this we calculated the percent
change between years for all of the data. If the indicators and employment figures of a sector are
closely correlated, then significant percent changes in the indicator should, either immediately or
after a certain interval of time, consistently induce a similar percent change in the number of people
employed in the sector. Otherwise, it can be said that there may be little or no correlation between
the two variables.
The two sets of percent change data for a sector were calculated in Microsoft Excel, and input
simultaneously into a smooth-line scatter plot. In this plot, the x-axis represented the year that each
data point came from, and the y-axis represented the magnitude of percent change of the data during
that year from the previous year. These graphs were first analyzed visually to determine if
Team ID #119, Page 6 of 18
employment changes closely followed indicator changes. However, this is a very qualitative way of
extracting data from a graph, and thus we turned to statistics to find a measure of correlation
between the data.
The coefficient of determination, or r2, is a standard statistical indicator of the amount of
correlation between two sets of data. The term can range from 0.0 to 1.0; a value of exactly 1.0
means that one set of data can be exactly mapped to the other via a linear function – one in the form
, where x values are taken from one set of data and y values from the other. In the
of
case of our economic model, a high r2 value for a job sector means that employment rates can be
closely modeled as a linear function of the sector’s indicator, whereas a lower r2 value indicates job
growth that is largely independent of stimulus factors.
Calculating one r2 value for each sector is not enough, however, because employment
growth may lag behind indicator growth by one or more years. This type of relationship is very
important to our analysis, yet would not be seen as a correlation by the standard r2 calculation.
Therefore, it was necessary to manually offset the graph of job growth by a variable amount of
years to account and check for a possible lag in market effects. This was done by writing a Java
program to calculate an r2 value for every possible shift.
The program took the two sets of percent change data as input, and, starting with the normal
mapping of indicator data to job growth data, iterated through calculating each possible r2 value.
Upon each successive iteration, each piece of indicator data (considered to be the independent
variable of the mapping) was chosen to map to the next successive piece of employment data
(considered the dependent variable, though it was sometimes the case that employment was
independent of the indicator). For example, in the original mapping, indicator data for the year 2000
was mapped to employment data for the year 2000; in the next iteration that indicator data was
mapped to 2001’s employment data, and in the iteration after that the indicator data was mapped to
2002’s employment data; this was done for every single piece of data (within the bounds of the
collected data) upon each iteration, and each iteration yielded a unique r2 value using the formula
where
,
are the averages of the two sets of data in the domain, range of the mapping, and
is the standard deviation of the indicator values in the domain. (sy,
defined similarly, is the standard deviation of the employment figures in the range.)
In each job sector, the local maximum r2 values were found by the Java program, and output
to a text file, along with the horizontal displacements that yielded them. Due to the small amount of
data available, we assumed that large lags were impossible to support, and therefore only focused
on the maximum r2 values that occurred after a minimal displacement. The end result was an
accurate factor of correlation between market indicators and employment, and a proposed lag
between indicator changes and employment changes.
Knowing how we could then link employment to indicators, we could then find the LeastSquares Regression Line (also known as LSRL or line of best fit). With these lines we could then
apply the conditions that the stimulus package would produce upon the market and get a rough
estimate of how employment in each sector of the market should end up.
Team ID #119, Page 7 of 18
VIII. Unemployment and Stimulus Package
A. Energy
a. Introduction
Energy, in particular renewable energy sources such as solar, wind, and nuclear energy, has enjoyed
immense media coverage due to a host of economic, environmental, and geopolitical concerns. In
devoting $18.5 billion by congressional authority to “Energy and Water,” Congress had made clear
that energy is a key component of the US economy, particularly with regards to consumer
confidence and industrial production. Energy spending occupied 3.5% of the $526 billion allocated
to be spent prior to 2010.
b.Assumptions
As per our methodology, we had to assume one particular representative indicator in the energy
sector as being a sensitive indicator of stimulus spending on that industry, once the money had been
spent. The energy sector was divided into two subsectors in our analysis: the oil and other
petroleum products industry and the renewable energy sources industry. We felt this division was
necessary due to the geopolitical implications, long-term surge in global demand, and popularity
among speculators commonly associated with the oil industry as well as the political and
environmental concerns often associated with renewable energies.
For the oil and petroleum products industry, US oil and petroleum products produced was assumed
to be the representative indicator of the industry, as it was the EIA’s (Energy Information
Administration’s) primary approximate indicator of overall consumption in the industry. For the
renewable energies industry, the primary indicator was assumed to be the private consumption of
solar and photovoltaic energy in the United States, due to the relatively nascent and minute nature
of the industry characteristic of renewable energy industries. The annual percent change of these
data series were tracked over time, juxtaposed with annual percent change in employment in that
sector.
c.The Model
0.2
U.S. Total Crude Oil and Petroleum Products Product Supplied versus Oil
and Gas Extraction Employment
Annual Percent Change
0.15
0.1
0.05
0
1975
-0.05
1980
1985
1990
1995
2000
-0.1
-0.15
U.S. Total Crude Oil and Petroleum Products Product Supplied
Oil and Gas Extraction Industry Employ ment
2005
2010
Team ID #119, Page 8 of 18
As seen in the graphs above, the peaks and troughs of each data series, for the most part, coincide.
With the exception of 1990, when the relationship became inverted, the two graphs generally
display a recognizable trend. The optimized coefficient of determination, obtained by making
variable the lag between the two curves, was found to be .222. This is due to the anomalies seen
between 1990 and 2000 to 2005. It is also important to note that very small variations in the supply
of crude oil had a greatly amplified impact on employment figures in the industry.
Solar/PV Energy Consumption versus Sector Employment
0.500
Annual Percent Change
0.400
0.300
0.200
0.100
0.000
1996
-0.100
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
Solar/PV Energy Consumption
Solar/PV Sector Employ ment
Despite an apparent lack of correlation to the naked eye, tracking consumption and employment
in the solar and photovoltaic energy sector yielded an extremely coefficient of determination, and
thus indicated a very high correlation. The coefficient of determination between these two data
series was .956. This high correlation appears misleading because of the large difference in
amplitude of the two data series. Like the previous graph displaying consumption of petroleum
products and employment in the petroleum industry, this graph shows that small fluctuations in
energy consumption lead to very large fluctuations in the sector’s employment.
d.Conclusion
Oil Industry Employment
The coefficient of determination for employment in the oil sector was too low (.222) to have a
reliable enough correlation to make predictions from.
Renewable Energy Employment
The renewable energy sector had a coefficient of determination of .955, our tightest fit, after
removing the outliers for 2001 (affected by a terrorist attack) and 2003 (affected by a hurricane
much more catastrophic than usual).
Based on the previous years of the construction sector, we predict that its change in employment, y,
can be predicted by the change in amount of money, x, in the sector by the relation:
This gave a predicted increase in employment of 63.5%.
2007
Team ID #119, Page 9 of 18
B. Taxes
a. Introduction
Perhaps exacerbated by political debate in mainstream US politics, tax policy is an extremely
contentious topic in US economic policy. However, because of historical precedents and intuitive
economic theory supporting government tax cuts in times of economic hardship, Congress was
successful in allocating 33% of the $526 billion of the American Recovery and Reinvestment Act to
be spent by 2010 on tax reductions.
We believe tax cuts to be the most effective measure of economic stimulus due to its reliance on the
free market to allocate the actual money spent, as well as the absence of a minimum one year lag
typically attributed to fiscal spending. This position will be elaborated upon in greater detail in
section VIII.
b.Assumptions
Once again, as per our methodology, we had to assume a particular representative indicator of the
impact of government tax policy on the economy once the money had been spent. This impact was
measured via the impact of government tax policy on private consumption, which being 70% of
GDP, we felt was an appropriate representative of the overall economy. In particular, the impact of
tax policy on private consumption was measured by tracking net national disposable income per
capita. This was calculated from the national disposable income by the BEA (Bureau of Economic
Analysis) and measured the portion of national income that is not allocated to taxes, savings, or
investments. Disposable income is a accurate measure of the impact of government tax policy on
private spending, as the quantity of disposable income per capita directly rises or declines in
response to a tax cut or tax raise, respectively.
c.The Model
Not surprisingly, tracking net national disposable income with overall national employment figures
painted a very visually convincing relationship between the two indicators. This is a good example
of clear leading and lagging indicators that, when tested for correlation via statistical analysis,
betrays the key relationships embedded in a visual examination. Every trough and peak seen in net
national disposable income can be linked to a corresponding trough and peak in national
employment, yet because of constantly varying periods and amplitudes, a statistical analysis yielded
an optimum coefficient of determination of .698.
Team ID #119, Page 10 of 18
Net National Disposable Income versus Employment
6.0%
Annual Percent Change
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
-1.0%1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
-2.0%
Net National Disposable Income
US Nonfarm Employment
d.Conclusion
Our modeling simulation shows that only an 8.4 % increase in employment should result from the
tax cuts offered by this stimulus package. The .72 coefficient of determination for this model gives
us strong confidence that this will be a very accurate prediction of the aid of this plan. This is based
on a total tax rebate of $173.6 billion dollars, distributed across a workforce 138 million people,
receiving $1257 each.
C. Housing and Construction
a. Introduction
Housing
To evaluate the impact of the stimulus package on the housing market, we investigated the
relationship between housing starts and the housing sector unemployment rate. Housing starts
record the number of new homes being built each month, as calculated by the Census Bureau and
the Department of Commerce. Housing (as a sector) is very accurate in predicting the future
direction of an economy considering the various components that go into building a home. Among
others, contractors, builders, designers, engineers, lawyers, and bankers are affected by home sales.
We hypothesized that more housing starts would correlate with increased employment in the
“housing industry.” To measure the historical levels of employment in the housing industry, we
utilized “residential building employment” as defined and published by the BLS (Bureau of Labor
Statistics).
Construction
Construction spending, published monthly by the Census Bureau, tallies up all spending on public
and private construction on both residential and nonresidential projects. The stimulus package
makes provisions for local and state governments to commence public works projects such as the
building of schools, roads, and highways. We compared the Census Bureau figures with the
“construction employment” figures published by the BLS in search of a relationship.
Overall
The stimulus package provided roughly $59.5 billion in 2009 and 2010, or 11.3% of all stimulus
funds spent in this period, for housing and construction in the public and private sectors.
Team ID #119, Page 11 of 18
b. Assumptions
Our major, well-founded, assumption is that increased stimulus provisions for housing will increase
housing starts figures, just as increased funds in the construction sector will encourage construction
spending.
c. The Model
Housing Starts versus Residential Building Sector Employment
30.0%
Annual Percent Change
20.0%
10.0%
0.0%
1980
-10.0%
1985
1990
1995
2000
2005
2010
2015
-20.0%
-30.0%
-40.0%
-50.0%
-60.0%
Housing Starts
Residential Building Sector Employment
As shown in the graph above, variations in housing starts tended to precede similar changes in
industry employment. With the exception of 2000, this trend held up since 1980. The optimized
coefficient of determination, obtained by making variable the lag between the two curves, was
found to be .662.
Construction Spending versus Construction Sector Employment
0.14
Annual Percent Change
0.12
0.1
0.08
0.06
0.04
0.02
0
-0.021992
1994
1996
1998
2000
2002
2004
2006
2008
-0.04
-0.06
-0.08
Construction Spending
Construction Sector Employ ment
As shown in the graph above, the peaks and trophs of the two data series overlap, with the
exception of the period between 1994 and 1996 where the curves became inverted. 2002 is a
2010
Team ID #119, Page 12 of 18
excellent reference point for this clearly distinguishable trend where decreases in construction
spending mimics decreases in construction sector employment. At some points, such as in 1997,
large decreases in construction spending resulted in very minimal reductions in employment. The
optimized coefficient of determination, obtained by making variable the lag between the two curves,
was found to be .748.
d. Conclusion
Construction Employment
The construction sector exhibited a coefficient of determination of .748.
Based on the previous years of the construction sector, we predict that its change in employment, y,
can be predicted by the change in amount of money, x, in the sector by the relation:
; with x and y in percent.
This gives a predicted increase in employment of 33.93%.
Housing Employment
The housing sector had a coefficient of determination of .638.
Based on the previous years of the housing sector, we predict that its change in employment, y, can
be predicted by the change in amount of money, x, in the sector by the relation:
This gives a predicted increase in employment of 7.82%.
As a special note, this sector exhibits a 1-year lag, so the effects predicted by this model are not
likely to be seen until 2011.
D. Labor, Health, and Education
a. Introduction
The stimulus package devotes approximately 30.1% ($161.8 billion) of all spending in 2009 and
2010 to education and health services. In order to forecast the effect these funds will have on
employment in this sector, we utilized the Healthcare SPDR which is a stock measuring an
amalgam of firms in the biotechnology, healthcare, and pharmaceuticals industries. Due to the
political pressure for healthcare reform, it comes as no surprise that a large portion of the stimulus is
devoted to this field. We investigated the historical relationship between increases in the value of
the SPDR and increases in employment in the healthcare industry.
b.Assumptions
We based our analysis on the assumption that increases in the Healthcare SPDR will continue to
mimic increases in healthcare employment in the future.
Team ID #119, Page 13 of 18
c.The Model
Health Care SPDR Closing Value versus Health Services Sector Employment
Annual Percent Change
0.2
0.15
0.1
0.05
0
1998
-0.05
2000
2002
2004
2006
2008
2010
-0.1
Health Care SPDR Closing Value
Health Services Sector Employment
Analysis on the healthcare sector was the most disappointing, especially given that 7.27% of the
$526 billion to be spent by 2010 of the American Recovery and Reinvestment Act was allocated to
the healthcare section. Again, as with the analysis on disposable income and national employment,
the lesson to be learned here is that a statistical correlation analysis will often belie the significant
relationship that is evident from an even cursory glance. Clearly, from this graph, there is absolutely
no relationship between the health care SPDR closing value and employment in the same sector.
However, the optimized coefficient of determination was .622, remarkably high given the clear lack
of correlation between the indicators evident from a visual examination.
d.Conclusion
Health care has been seen to be invariant with changes in GDP and consumption in the health
sector.
The only trend to be seen is a steady acyclic rise over time.
The graph of GDP and health services illustrates these facts.
IX. Further Stimulus
Without further stimulus, the Survey of Professional Forecasters 2009Q1 estimates that the US
economy will experience a 2.0% contraction over the coming year. The economy would continue to
be flat until late 2009. This represents an unacceptable decline in real wealth. Christina Romer, the
Chair of the Council of Economic Advisers, suggested in a recent paper that for every $1 of
exogenous tax changes, the GDP will experience a change of $3 in the same direction. This tax
multiplier is distinct from the Keynesian spending multiplier we rejected earlier. The paper
estimates that the effects of the new tax code will be long lasting and that the change will be most
fully realized after 10 months. We elected to eliminate the corporate income tax to stimulate the
economy. This represents a sector-blind method of stimulating the economy. It avoids sponsoring or
supporting the special priorities of either political party. Putting extra money in corporate coffers
Team ID #119, Page 14 of 18
will ease the strain on their balance sheets and hopefully stem the job loss hemorrhage. Healthier
corporate finances translate into greater hiring power. The IRS collected $312.1 billion in corporate
income tax returns in 2005. Eliminating this tax would not make the federal budget much more
unbalanced than it already is.
For this model, we will work in 2000 chained dollars to account for inflation. Thus, income tax
returns in 2005 amounted to $253.3 billion dollars. Assuming that one-fourth of that amount is
conserved by corporations each year because of the tax repeal, three-fourths of the amount should
be added to GDP every quarter. Assuming the stimulus would be passed almost immediately, the
second quarter would be the first to be affected. Thus,
where GDPn is the GDP in the nth quarter, n goes from 2009Q2 to 2010Q1, and r is the predicted
percent change. The resulting improvement in the real GDP over the nonstimulated baseline is
substantial. Without the stimulus, the economy is predicted to shrink 2.0%. With the stimulus, it is
predicted to grow 4.8%.
Real GDP
12,200.0
12,000.0
Billions of $
11,800.0
11,600.0
11,400.0
11,200.0
11,000.0
Baseline
10,800.0
10,600.0
2008.50
With Stimulus
2008.75
2009.00
2009.25
2009.50
2009.75
2010.00
2010.25
Tim e (Fiscal Quarters)
Quarter
2008Q4
2009Q1
2009Q2
2009Q3
2009Q4
2010Q1
Real Baseline
% Quarterly
Real Stimulated
% Quarterly
GDP (billions) Change
GDP (billions)
Change
11,525.00
11525.00
10925.70
-5.2
10925.70
-5.2
10729.04
-1.8
10919.01
-0.06
10836.33
1.0
11218.18
2.7
11031.38
1.8
11610.08
3.5
11296.13
2.4
12078.70
4.0
It is generally accepted that increases in real GDP are reflected by a decrease in unemployment.
Historically, this is apparent from simple graphical analysis.
Team ID #119, Page 15 of 18
Unemployment Lagging Real GDP
0.1
0.08
0.06
0.04
0.02
0
-0.021960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
-0.04
-0.06
-0.08
-0.1
-0.12
Unemployment
Real GDP
Thus, eliminating the corporate income tax propels the economy towards robust growth in the
coming year.
X. Conclusion
To recapitulate, the sectors impacted by the stimulus package were energy, health and
education, construction (including transportation), as well as an analysis on tax cuts. Furthermore,
the energy sector was subdivided into petroleum and renewable energy sectors. Thus six different
sets of data were ultimately analyzed. The coefficients of determination are displayed in the
following table and chart:
Optimized R2 Values
Construction Employment Versus Construction
Spending
Gas Extraction Employment Versus Petroleum
Supplied
House Building Employment Versus Housing
Starts
National Employment Versus Disposable Income
Health Employment Versus Health Care Value
Employment Versus Energy Consumption
Best Correlation / R2 Value
0.748
0.222
0.662
0.698
0.622
0.956
2015
Team ID #119, Page 16 of 18
Optimized R2 Values Between Sector Indicator and
Employment
1.200
1.000
0.800
0.600
0.400
0.200
0.000
Gas Extraction
House Building
National
Health Employment Employment Versus
Construction
Employment Versus Employment Versus Employment Versus Employment Versus Versus Health Care Energy Consumption
Petroleum Supplied:
Housing Starts:
Disposable Income:
Value:
Construction
Spending:
Furthermore, the relationships between the representative indicators and employment figures in
each sector were extrapolated to calculate an expected percent increase in employment in each
sector for the year 2010 and included in the conclusion section for each sector. The following table
summarizes these statistics:
Industry Name
Expected % Increase in Employment in 2010
Housing
Renewable Energy
Construction (excluding Housing)
7.82%
63.5%
33.93%
By our analysis, spending and investment in renewable energy proved to be the efficient allocation
of the stimulus bill, followed by housing. These results are not surprising as renewable energy is a
very nascent industry with a high potential for growth, and the housing market has been severely
depressed with the crashing of the housing bubble. The petroleum and health care industries were
deemed to be of no statistical significance in terms of correlation; the petroleum industry because of
a low coefficient of determination and the health care industry because the employment trends in
the sector are clearly acyclical.
Testing
This model has tremendous potential as an application for economists, academics, individuals,
corporate executives, and politicians with a vested interested in the outcome of the economy and the
impact of the stimulus bill. Because of our heavy dependence on empirical methods and our
independence of hotly disputed Keynesian multipliers, we are strongly confident that our model will
be highly effective in predicting the future impact of government spending and tax cuts on the
economy, especially sector by sector. The emphasis on individual sectors in the economy is what
we believe to be our model’s strongest point, as the economy is simply too large to be analyzed as a
whole, and our model allows the broadness of the economy to be analyzed in a practical and
facilitative manner.
Team ID #119, Page 17 of 18
XI. Works Cited
United States. Internal Revenue Service. Corporation Income Tax Returns. Washington, D.C.:
GPO, 2005.
Romer, Christina D. and David H. Romer. "The Macroeconomic Effects of Tax Changes: Estimates
Based on a New Measure of Fiscal Shocks." Nov 2008.
United States. Federal Reserve Board. Summary of Commentary on Current Economic Conditions .
Washington, D.C.: 2009.
Miron, Jeffery A.. "Economic Change We Can Believe In." [Weblog Reasononline] 06 Feb 2009.
Reason Foundation. 8 Mar 2009 <http://www.reason.com/news/show/131556.html>.
United States. Federal Reserve Bank of St. Louis. Series: USEHS, All Employees: Education &
Health Services. St. Louis: 2009.
United States. Congress. American Recovery and Reinvestment Act of 2009. Washington, D.C.:
GPO, 2009.
United States. Bureau of Economic Analysis. National Economic Accounts. Washington, D.C.:
GPO, 2009.
United States. Census Bureau. New Residential Construction. Washington, D.C.: 2009.
United States. Energy Information Administration. Annual Solar Thermal Collector Manufacturers
Survey. 2006.
Eggertsson, Gauti B.. United States. Federal Reserve Bank of New York. Fiscal Multipliers and
Policy Coordination. New York: 2006.
Baumohl, Bernard. The Secrets of Economic Indicators, Hidden Clues to Future Economic Trends
and Investment Opportunities (2nd Edition) (Wharton School Publishing Paperbacks). Grand
Rapids: Wharton School, 2007.
Friedman, Milton. Capitalism and Freedom. Chicago: University of Chicago Press, 1962.
Bergman, U. Michael. "The Keynesian Multiplier." (2005)
Elmendorf, Douglas W. United States. Congressional Budget Office. Letter to Senator Judd Gregg.
2009.
United States. Congressional Budget Office. Cost Estimate: H.R. 1. Wasington, D.C.: GPO, 2009.
Elmendorf, Douglas W. United States. Congressional Budget Office. Letter to Senator Charles E.
Grassley. 2009.
Elmendorf, Douglas W. United States. Congressional Budget Office. Letter to Congressman Paul
Ryan. 2009.
Team ID #119, Page 18 of 18
Yourish, Karen. "Taking Apart the $819 billion Stimulus Package," Washington Post. 01 Feb 2009.
8 Mar 2009. < http://www.washingtonpost.com/wpdyn/content/graphic/2009/02/01/GR2009020100154.html >.
United States. Bureau of Economic Analysis. Government Expenditures and Reciepts. 2009.
"Final Version H.R. 1." Read the Stimulus. 8 Mar 2009
<http://readthestimulus.org/index.php?doc=hr1final>.
United States. Bureau of Labor Statistics. 2009.
Yates, Daniel S. Practice of statistics TI-83/89 graphing calculator enhanced. New York: W.H.
Freeman, 2003.
Ellis, Joseph H. Ahead of the Curve. Boston: Harvard Business School Press, 2005.
"Health Care SPDR." Google Finance. Google, Inc.. 8 Mar 2009
<http://www.google.com/finance?q=XLV>.